def read(self, f_handle, f_id='a3m', remove_insert=True):
"""Read a sequence file
Parameters
----------
f_handle
Open file handle [read permissions]
f_id : str, optional
Unique sequence file identifier
remove_insert : bool, optional
Remove insert states [default: True]
Returns
-------
:obj:`SequenceFile <conkit.core.sequencefile.SequenceFile>`
"""
# Create a new sequence file instance
sequence_file = SequenceFile(f_id)
# Read any possible comments and store in file remarks
while True:
line = f_handle.readline().rstrip()
if not line:
continue
elif line.startswith('#'):
sequence_file.remark = line[1:]
elif line.startswith('>'):
break
# Read the sequence record(s) and store them
while True:
if not line.startswith('>'):
raise ValueError("Fasta record needs to start with '>'")
id = line[1:] # Header without '>'
chunks = []
line = f_handle.readline().rstrip()
while True:
if not line:
break
elif line.startswith('>'):
break
chunks.append(line)
line = f_handle.readline().rstrip()
seq_string = "".join(chunks) # Sequence from chunks
# Remove insert states
if remove_insert:
seq_string = self._remove_insert(seq_string)
# Create the sequence record instance
sequence_entry = Sequence(id, seq_string)
# Store the sequence in the file
try:
sequence_file.add(sequence_entry)
except ValueError:
while True:
new_id = sequence_entry.id + "_{0}".format(
np.random.randint(0, 100000))
if new_id in sequence_file:
continue
else:
break
sequence_entry.id = new_id
sequence_file.add(sequence_entry)
if not line:
break
# Match the insert states of the sequence
if not remove_insert:
self._adjust_insert(sequence_file)
return sequence_file
def SequenceFile(*args, **kwargs):
""":obj:`SequenceFile <conkit.core.SequenceFile.SequenceFile>` instance"""
from conkit.core.sequencefile import SequenceFile
return SequenceFile(*args, **kwargs)